public virtual void TestTextFromPdfLayers()
{
String testName = "testTextFromPdfLayers";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetImageLayerName("Image Layer");
properties.SetTextLayerName("Text Layer");
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(new CustomOcrEngine(), properties);
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
(pdfPath));
NUnit.Framework.Assert.IsNotNull(doc);
IList <PdfLayer> layers = doc.GetCatalog().GetOCProperties(true).GetLayers();
NUnit.Framework.Assert.AreEqual(2, layers.Count);
NUnit.Framework.Assert.AreEqual("Image Layer", layers[0].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[0].IsOn());
NUnit.Framework.Assert.AreEqual("Text Layer", layers[1].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[1].IsOn());
doc.Close();
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, PdfHelper.GetTextFromPdfLayer(pdfPath, "Text Layer"
));
NUnit.Framework.Assert.AreEqual("", PdfHelper.GetTextFromPdfLayer(pdfPath, "Image Layer"));
}
/// <summary>
/// Perform OCR using provided path to image (imgPath)
/// and save result PDF document to "pdfPath".
/// </summary>
/// <remarks>
/// Perform OCR using provided path to image (imgPath)
/// and save result PDF document to "pdfPath".
/// (Method is used for compare tool)
/// </remarks>
protected internal virtual void DoOcrAndSavePdfToPath(AbstractTesseract4OcrEngine tesseractReader, String
imgPath, String pdfPath, IList <String> languages, IList <String> fonts, Color color)
{
if (languages != null)
{
Tesseract4OcrEngineProperties properties = tesseractReader.GetTesseract4OcrEngineProperties();
properties.SetLanguages(languages);
tesseractReader.SetTesseract4OcrEngineProperties(properties);
}
OcrPdfCreatorProperties properties_1 = new OcrPdfCreatorProperties();
properties_1.SetPdfLang("en-US");
properties_1.SetTitle("");
if (fonts != null && fonts.Count > 0)
{
FontProvider fontProvider = new FontProvider();
foreach (String fontPath in fonts)
{
String name = FONT_PATH_TO_FONT_NAME_MAP.Get(fontPath);
fontProvider.GetFontSet().AddFont(fontPath, PdfEncodings.IDENTITY_H, name);
}
properties_1.SetFontProvider(fontProvider);
}
if (color != null)
{
properties_1.SetTextColor(color);
}
if (languages != null)
{
NUnit.Framework.Assert.AreEqual(languages.Count, tesseractReader.GetTesseract4OcrEngineProperties().GetLanguages
().Count);
}
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(tesseractReader, properties_1);
try {
using (PdfWriter pdfWriter = GetPdfWriter(pdfPath)) {
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(new FileInfo(imgPath
)), pdfWriter);
NUnit.Framework.Assert.IsNotNull(doc);
doc.Close();
}
}
catch (System.IO.IOException e) {
LOGGER.Error(e.Message);
}
}
public virtual void TestOriginalSizeScaleMode()
{
String path = PdfHelper.GetDefaultImagePath();
FileInfo file = new FileInfo(path);
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(new CustomOcrEngine());
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
());
NUnit.Framework.Assert.IsNotNull(doc);
ImageData imageData = ImageDataFactory.Create(file.FullName);
float imageWidth = GetPoints(imageData.GetWidth());
float imageHeight = GetPoints(imageData.GetHeight());
float realWidth = doc.GetFirstPage().GetPageSize().GetWidth();
float realHeight = doc.GetFirstPage().GetPageSize().GetHeight();
NUnit.Framework.Assert.AreEqual(imageWidth, realWidth, DELTA);
NUnit.Framework.Assert.AreEqual(imageHeight, realHeight, DELTA);
doc.Close();
}
public virtual void TestPdfLayersWithTextLayerOnly()
{
String path = PdfHelper.GetDefaultImagePath();
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetTextLayerName("Text Layer");
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(new CustomOcrEngine(), properties);
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
());
NUnit.Framework.Assert.IsNotNull(doc);
IList <PdfLayer> layers = doc.GetCatalog().GetOCProperties(true).GetLayers();
NUnit.Framework.Assert.AreEqual(1, layers.Count);
NUnit.Framework.Assert.AreEqual("Text Layer", layers[0].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[0].IsOn());
doc.Close();
}
public virtual void TestPdfLayersWithDefaultNames()
{
String path = PdfHelper.GetDefaultImagePath();
FileInfo file = new FileInfo(path);
OcrEngineProperties ocrEngineProperties = new OcrEngineProperties();
ocrEngineProperties.SetLanguages(JavaCollectionsUtil.SingletonList <String>("eng"));
CustomOcrEngine engine = new CustomOcrEngine(ocrEngineProperties);
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(engine);
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
());
NUnit.Framework.Assert.IsNotNull(doc);
IList <PdfLayer> layers = doc.GetCatalog().GetOCProperties(true).GetLayers();
NUnit.Framework.Assert.AreEqual(0, layers.Count);
doc.Close();
NUnit.Framework.Assert.AreEqual(engine, ocrPdfCreator.GetOcrEngine());
NUnit.Framework.Assert.AreEqual(1, engine.GetOcrEngineProperties().GetLanguages().Count);
NUnit.Framework.Assert.AreEqual("eng", engine.GetOcrEngineProperties().GetLanguages()[0]);
}